Will Cox: Writing Online

Earth to John and Dave: RTFM

Wed, 27 Aug 2003 16:31:00 GMT

Dave notes that UserLand's RSS PageRank is now zero. Google doesn't follow redirects when calculating PageRank. My old PageRank was 7/10. Now it is zip. I have also found that Google isn't updating the calclation that often. They must be batching it each quarter. I have seen lots of sites stuck at zero even though they are popular. [John Robb's Weblog]

Google does follow redirects. However, and here's where you're losing it, for PageRank to follow they have to be 301 Moved Permanently redirects. Google couldn't give a shit about Dave's attempt at an RSS redirect, and 302 Found redirects are temporary, so why should the index update anything?

The You RI

Fri, 11 Jul 2003 15:43:38 GMT

Rick Klau notes that John Robb site at UserLand left immediately after John himself did. There is an object lesson in this:

Final lesson from this situation: never, ever blog at a domain that’s not owned by you. Don’t blog on your employer’s site, don’t blog on your blogging application’s site — make sure your blog lives (and stays!) on your own domain.

The things of this world are short-lived, die and rot. But be prudent and plan to take your stuff with you when you go. Technically that means use an e-mail address where both sides of the @ are yours. When you use someone else's site, whether pages.prodigy.net or radio.weblogs.com, you're renting. Eventually, you'll want to move out of your parents' house and stake a claim of your own.

But I thought that's what HREFs were for!

Tue, 01 Jul 2003 18:29:22 GMT

I'm really bothered by this exchange over link and guid in the rss2 specification. (I made a comment, but apparently I wasn't scatological enough.)

Within the channel element, link is "The URL to the HTML website corresponding to the channel."

Within the item element, link is The URL of the item.

Within the item element, guid is A string that uniquely identifies the item.

What is an item?

An item may represent a "story" — much like a story in a newspaper or magazine; if so its description is a synopsis of the story, and the link points to the full story. An item may also be complete in itself, if so, the description contains the text (entity-encoded HTML is allowed), and the link and title may be omitted. All elements of an item are optional, however at least one of title or description must be present.

[INSERT MORE HERE ap example, nyt example, me]

But then we have this clarification: link should be used only to link to the article being described by the post. Here's an example. In the example you'll find an item with 1) a title, 2) a link, 3) a description of the link, and 4) a guid, which is not a "permalink." Why? Because

If the guid element has an attribute named "isPermaLink" with a value of true, the reader may assume that it is a permalink to the item, that is, a url that can be opened in a Web browser, that points to the full item described by the element.

So, since thing described by my item element is not this, there is no permalink? That's a bit confusing.

And what it boils down to is that some vociferous people find it obscenely difficult to communicate effectively.

More Wood for the Fire

Mon, 30 Jun 2003 19:45:46 GMT

This comment is not productive, and, since I don't care to find the documents backing my memory, may be wrong.

In 2001, the UserLand RSS version was 0.92, and all elements of item were made optional. Specifically, title and link became optional. This is the version of RSS that I became acquainted with from using Radio. The earliest sample I have is this one from April, 2002.

On March 12, 2002, the Radio application was enhanced to allow users to specify a title and a link for each item. Prior to that enhancement, users could specify only the description (and category); everything else was programmatically generated.

The link element that users specified was not necessarily the same as the URI for the HTML rendering of the item. It could be any URI, though in practice it was either the URI of the item's rendering, or of some thing referenced by the text in the description.

Radio stuffed this user-specified link into the RSS link element, and the title into the RSS title element. In rendering the HTML version, the title was styled, and the link added. Here is a sample 0.92 file from June, 2002. Some items have no title, while others do. Each item has a link.

This change to Radio caused a problem. Before, the aggregator could assume that the link element referred to the HTML version, and could act in a particular fashion based on that assumption. Making the link element user-editable invalidated this assumption.

This could be resolved in two ways. First, Radio could use the specified link only in the HTML rendering, while using the HTML URI in the RSS link element. Alternately, realizing the need for a means of identifying the item itself, the RSS specification could be revised.

The second option was chosen. This additional element is the guid.

The Replacements

Thu, 10 Apr 2003 13:15:54 GMT

Bundle Option 1:

Optimal, Improper Redirection

Wed, 09 Apr 2003 19:30:21 GMT

Yesterday I wrote about a means of causing two RSS feed consumers to update their subscriptions list, and about a means of causing web browsers to load another page. This is not the best way to do either of those things. However, it may be the optimal way.

The optimal way is that which works best given certain constraints. In this situation the constraint is two-fold:

end-users are unable to modify the server configuration (by placing directives within a .htaccess file)
This is the closest thing to a meta refresh workaround for RSS aggregators.

There are downsides to these methods. The downsides of the meta refresh element are that

it is bandwidth intensive, and
machines do not update themselves

By the latter I mean those like our friend Google, which will not update its index, and so PageRank will be lost. Some machines simply do not follow that element, and are stuck wondering where you went.

The downsides of the attempt to redirect RSS are many.

It's not RSS.
Only Radio Userland and NetNewsWire do anything with that file, so
people who read your feed in other aggregators will never know you moved.
It's a machine-readable format that means nothing to machines, and is barely comprehensible for humans.
It requires extra effort on your part.

So what should you do? If your readership only uses Radio Userland and NetNewsWire, then that fragment of XML is fine. If they don't, then a politely worded note in the RSS must suffice — perhaps in combination with a polite request to Userland to be considerate of others.

Ch-ch-changes

Wed, 09 Apr 2003 00:29:01 GMT

modified system.verbs.builtins.radio.weblog.render (called from macros.viewWebLog) to remove some items that should properly be in the #itemTemplate.txt file.

in itemsText, empty anchor (a) element at start of item
in the bundle which sets itemTitle, assigning a class to the weblogItemTitle is nice and all, but don't you think that shouldn't be in the script?
system.verbs.builtins.radio.data.website.["#tools"].listCategoriesForPost, is compiled code, so I can't edit it. However the div enclosing the category list should be removed, allowing that list to be formatted in the #itemTemplate.txt file

also made some changes to the #itemTemplate.txt file

removed extra   and <br /> elements.
the span enclosing the datestamp should be a paragraph

Reminder: Your Subscription Will Expire

Tue, 08 Apr 2003 14:11:12 GMT

Userland provides site hosting and updates for Radio Userland users on a subscription basis. The cost of the first year of this subscription is included in the price of Radio. If the subscription is not renewed, then two things will happen:

The Radio Userland program will no longer update itself.
Files on radio.weblogs.com will be removed.

If you decide not to renew, and will continue publishing from another location, there are some things you should do to prepare the world at large, to notify your readers of the change. Readers come in all shapes and sizes. Some of them are human, others are machines. Humans, as a rule, are somewhat more flexible than machines, and can be notified in a plethora of ways. Machines like things to be just so. Not only do you have to speak their language, but you have to be grammatically correct.

Since we're speaking of a web site here, the machines in question speak the Hypertext Transfer Protocol (HTTP), and may understand HTML and RSS. This is good, because you won't be able to use an HTTP Redirect on radio.weblogs.com, with all that implies.

For the HTML version of your site, edit your template to include a meta element, then re-publish everything. This will cause some machines to load the new location. This is an example that will cause the browser to load the new location zero (0) seconds after loading the original location.

<meta http-equiv="refresh" content="0;url=http://weblog.bluepenguin.us/0106188/index.html">

Joe Friend passes on a UserTalk macro which will customize the meta element for each page:

<meta http-equiv="refresh" content="5;url=http://www.newsite.com/<%local (pta = html.getpagetableaddress ()); local (path = pta^.path); return (string.replace (path, ".txt", ".html"))%>">

Some machines, primarily search engines, do not pay attention to this element.

For the RSS version, there was some argument a while back about a method for using RSS to inform feed consumers of a change in the status of an URL. There was no real agreement on what to do about this, but discussion died once Dave Winer wrote something for Radio Userland. Due to the lack of consensus, this has only been implemented in Radio and NetNewsWire. This does not reduce its utility, since at least a subset of your readers will be moved to follow.

The following procedure is not intended to be a comprehensive guide to site migration. While it assumes that you will not use Radio to update your non-Userland-hosted site, the differences between this and simply using another host for your Radio-generated site are minimal.

Edit your template to include the meta element mentioned above.
Write a farewall post for the Userland-hosted site.

Hand edit your rss.xml file to point to the new location.


<?xml version="1.0"?>
<redirect>
<newLocation>http://weblog.bluepenguin.us/0106188/index.xml<;/newLocation>
</redirect>

Upstream everything to radio.weblogs.com.
Turn off the Radio.

Any questions? This method is not recommended if your readers do not use Radio Userland.

Reformatting the Navigator Links

Thu, 03 Apr 2003 14:54:11 GMT

In Radio Userland, you can maintain a list of links you like, and Radio will automatically insert them into each page. These are called Navigator Links. They are stored by default as an XML file, #navigatorLinks.xml, and placed into the template with the variable <%navigatorLinks>. While rendering HTML from the template, Radio calls system.verbs.builtins.radio.html.drawNavigatorLinks in radio.root.

You may not wish to edit drawNavigatorLinks, but you may wish to remove some of the cruft added during the rendering of the file. drawNavigatorLinks behaves differently depending on what kind of file it's rendering. text/xml files are parsed and modified, as are text/opml files, though in different ways. Any other kind of file is included as is.

How is the file type determined? By looking at the extension, of course! So, move the file #navigatorLinks.xml to #navigatorLinks.txt and format it however you'd like, remembering that it will be displayed within an HTML file.

Radio Radio

Wed, 19 Mar 2003 14:23:53 GMT

Got my renewal notice in the e-mail the other day. (Hi, Jake.) I have decided not to renew. I looked at EditThisPage in January of 2000, then briefly used it, jogger and Blogger in 2001, once I decided that I may as well inflict my opinions on the world instead of just my friends. Both EditThisPage and Blogger quickly convinced me that I like to write off-line, so I tried Radio. The aggregator hooked me. The inability to make simple changes has driven me away. So now I'm looking for two things: a passable aggregator, and time to move my archives.

And Rot Inside a Corpse Shell

Thu, 16 Jan 2003 16:19:45 GMT

I, being the slow sort, just realized something about fragment identifiers. HTML 4.01 has two types of fragment identifiers: id and name attributes. So does XHTML 1.0, but there name is deprecated. So, assuming that you wish to move from HTML to XHTML, you will choose to use id instead of name. However, the range of values of these attributes is different, because of their differing datatypes. What does this mean? This means that during the course of a conversion from HTML to XHTML, some URIs will break — unless steps are taken elsewhere, outside of the document, to prevent such breakage.

Impossible Things

Fri, 10 Jan 2003 19:19:45 GMT

One of the things that I learned from Mark Pilgrim the other day was that the cite attribute of the blockquote and q elements is an URI. Do you know how hard it is to find an URI for a citation? I guess the folks at CiteSeer would.

`I ca'n't believe that!' said Alice. `Ca'n't you?' the Queen said in a pitying tone. `Try again: draw a long breath, and shut your eyes.' Alice laughed. `There's no use trying,' she said `one ca'n't believe impossible things.' `I daresay you haven't had much practice,' said the Queen. `When I was your age, I always did it for half-an-hour a day. Why, sometimes I've believed as many as six impossible things before breakfast. There goes the shawl again!'

IE doesn't render the q appropriately; it should. Weblog tools should, when posting a quote, automatically complete the cite attribute. But I suppose that neither of those things will happen.

Save Me!

Wed, 08 Jan 2003 16:54:37 GMT

Don't you just hate it when you've got a post almost finished, and you go looking for that last link to top it off, and poof! there it goes into the bitbucket.

Samuel Pepys

Fri, 27 Dec 2002 14:11:51 GMT

This on-line rendition of Samuel Pepys's Diaries by Phil Gyford is simply incredible.

Keeping Up With the Joneses

Mon, 23 Dec 2002 18:37:27 GMT

L. M. Orchard has released a MoveableType plug-in to allow the use of an LDAP directory for authenticating authors. This is an essential step in integrating MoveableType into an existing infrastructure. Duplication and attempted synchronization of authentication credentials and user information generally degrades the user experience and introduces errors. Supporting a common access protocol enables the introduction of new services while keeping the costs of deployment low.

Make a Wish

Thu, 19 Dec 2002 18:34:28 GMT

Just ran across on this in my referrer log.

Cool URIs Require Forethought

Tue, 10 Dec 2002 19:04:07 GMT

I've moved my journal several times. I experimented with EditThisPage, with Blogger, and then with Radio Userland. I've not tried to maintain continuity, because each was an experiment. Indeed, my Blogger site is now 410 GONE. I've changed hosts from BestWeb to Pair after I bought my own domain, and in the course of doing so changed the underlying directory from radio to journal. And now, I'm changing it from journal to journal.

This change will not invalidate the previous URI. I've created a symbolic link between journal and journal, so as far as the operating system is concerned, those directories are the same. I'm also rewriting the URL. Tim Berners-Lee discussed why cool URIs don't change, and while there are a number of means for maintaining deprecated URIs, the important thing is planning a naming scheme. Cool URIs require forethought.

That doesn't happen.

Why Does Dave Winer Hate Ones?

Tue, 10 Dec 2002 12:40:35 GMT

This is whack. My post yesterday on jumping to conclusions contained a string that caused Radio's news aggregator to render it incorrectly. In the third week of August I noticed something similar. This bug is just too stupid. Under normal circumstances I would try to narrow the test case in order to reproduce the bug consistently, so that it can be fixed, but down that road lies only frustration. See what it did to the title of this post?

Good to See You

Wed, 04 Dec 2002 13:51:30 GMT

Robert Shaw, of the International Telecommunications Union (ITU), is back on-line, with some excellent posts on DNS

Ignorance is No Excuse

Tue, 03 Dec 2002 20:02:08 GMT

Tool vendors have the great responsibility of ensuring that their work produces compliant code. Most of the time, all it takes is simple changes, like producing lowercase tags instead of UPPERCASE. Other times, it takes a small bit of knowledge, like knowing about fragment identifiers, and using them. But there's no excuse when help is offered, and not taken. (With the links there, this sounds spiteful. It's not. Just an observation about accepting patches.)

Linkages

Tue, 26 Nov 2002 20:21:13 GMT

Jon Udell -> AllConsuming -> Mullahs on the Mainframe -> unmedia -> subscribed I'm just too much of a dilettante to concentrate on system administration for longer than it takes to type reboot. :-)

Time for liveTopics?

Fri, 22 Nov 2002 16:11:30 GMT

%du -hs .
 85M    .
%du -hs journal/
 53M    journal/

Average data transferred per day: 35.726 megabytes

Feeping Creaturism

Thu, 21 Nov 2002 14:10:14 GMT

In the course of writing a warning to others, Mark Pilgrim notes some things I intend to try on my systems. But I wonder, some of the decisions made by the changes in XHTML 1.1 seem just plain stupid. In any case, I'm still peeved by editors that introduce excessive <p>'s and CMSes that have no excuse for not adequately separating the data from the presentation. I should not have to tweak the defaults to get my page to validate.

Rebirth

Fri, 01 Nov 2002 16:28:53 GMT

I'm changing the name here, to remove the word "gallery." Apparently that's a signal to some sick bastards that this is a pr0n site, which it's most emphatically not. Most of my hits from non-Google search engines are pr0n-related. And most of my Google hits are peanut-related. When I chose the name, I couldn't think of anything better. I still can't think of anything that's not already in use. But I do have an alternate spelling at my disposal ;-) Now I just have to think about how to merge Cox Crow with /journal, without destroying my googlejuice and my URIs. Cool URIs change all the time, for just this reason — doesn't mean that the links have to rot, though.

Beauty in Text

Tue, 22 Oct 2002 16:27:30 GMT

Inluminent pointed out Textism's Refer for tracking referrers using MySQL and PHP. Textism reminded me that I love typefaces, and haven't done enough to spruce up my site. I'm particularly fond of Garamond, Goudy Old Style and Gill Sans, though Caslon and Sabon are wonderful.